Gemini gave an incorrect response while I was looking up a Magic the Gathering ruling. I’m not sure if this is the right place to talk about it. I submitted feedback, but I want to post it here in case it doesn’t get read because, based on this result, I am sure there are several other similar cases.
With Magic the Gathering, I expected consistently accurate answers since it is apparently Turing-complete. Its online clients have the rules automated. Gemini clearly doesn’t have access to MTG’s ruleset and instead seems to rely solely on public online discussion for its responses. Maybe it’s a copyright issue? Should it be Google’s job to train AI to be as capable as a human expert with regards to MTG rulings, or should it be Wizards of the Coast’s job to produce an AI with human like MTG ruling skill? Until then, it is inadvisable to rely on AI for MTG rulings. It is better for now to use the standard search engine for results to reference.